The UC San Diego Police Department publishes daily Crime Logs to their website. This website displays a filtered subset of those Crime Logs.
Everyday at 8 AM, 12 PM, and 5 PM PST, the server will check for any new updates posted to the official UCSD crime log website and update this site accordingly.
< Incident Type >
< Location >
< Date >
< Time >
Summary:
< Crime Log Summary >Disposition:
< What police did about it >
Animal Call
Hopkins Parking Structure
9/20/2022
1:07 PM
Summary:
Report of a squirrel acting aggressive Disposition:
Checks OK
Machine learning. I used Scikit-learn to do the heavy lifting.
In a nutshell I:
- Manually labeled 2,341 individual crime logs as either "Interesting" or "Not Interesting" (true or false).
- Tokenized each crime log summary by stemming and removing stop words using the Python NLTK library.
- Utilized Scikit-Learn's TF-IDF vectorizer to extract features from the tokenized crime log summaries.
- Trained a Linear Support Vector Classifier to classify crime logs as “interesting” or “not interesting”.